Hello,

I'm building what could be called a hybrid between a music studio (for recording, mixing, and some live performances) and a space for listening/dancing in a single room (about 785 sqft).

I'm working with an acoustician, and we're planning to install a Dolby Atmos system in there (for a couple of reasons).

One of the use cases will be to perform live music, mostly using hardware (synths, reverbs) and software (VSTs via DAW), performed and mixed live in Dolby Atmos.

I'm a bit worried about the latency. I'm considering going the DANTE route with MTRX Studio (it seems to be the best bang for the buck option despite not using Pro Tools), and I can't really tell how much latency I can expect.

At the moment, I have a composer setup with what I would consider a powerful slave PC:
- AMD Ryzen 9 5950X, 16x 3.4GHz
- PNY Quadro RTX 4000, 8GB GDDR6
- 128 GB DDR4-3000 CL16 RAM,
- RME HDSPe AIO Pro

I use that slaved via Vienna Ensemble Pro over 10Gbe (as many composers to) to my Macbook Pro 2021 M1 Max, with about 600-800 tracks on Nuendo (for composing orchestral music) that get their audio from VEP.

I'd be willing to work in Live, Nuendo, Logic - even Pro Tools, if necessary.

At the moment I'm unsure or don't know the following:
1. How much latency can I expect, ideally?
2. What would be the lowest latency solution for live Dolby Atmos performances/mixing?
3. Is it better (in terms of latency and/or stability) to a) use the slaved PC for the Dolby Atmos renderer, b) use e.g. Nuendo's internal renderer, or c) use the renderer on the same machine as the DAW (which seems to be possible on Mac)?

I've worked my way through every Dolby Atmos video and article I could find, but didn't find straightforward answers to this anywhere, so any help and pointers are very welcome!

Thanks

The new version 5 renderer states the associated processing latency in the audio menu. From memory, the minimum is 22ms. On top of that you will have your DAW I/O and sample buffer latency, and the DAW to Renderer I/O latency.

The choice of DAW is probably a bit irrelevant as they all support the same hardware and sample buffer settings for audio interfaces. Integrated renderers might be a different question. I have never used the internal renderer in Nuendo, but it shouldn't be too hard to do your own round trip tests to figure out what the latency is. In theory it could be quite low as there are fewer trips through various kinds of audio IO. In practice, it might be buffered to hell and have much higher latency.

Beyond that, beware of virtual connections. Physical I/O like RME HDSPe MADI or Danteface cards will have low, stable latency. Dante DVS isn't terrible, but the latency is higher. Stuff like Dolby Audio Bridge and Pro Tools Audio Bridge is very high and also unpredictable between each playback.

Thanks for the info, that's a start. So if I have about 20ms total latency from daw (about 7-8ms from the buffer and probably 10-12 from plugins + hardware send/returns) and add 22ms for atmos rendering, that's about 40-50ms minimum latency from moving a knob to the audience hearing the effect. That's similar to having a distance of 17m/55ft to someone.

It sounds like a lot, but I'm unsure. Any thoughts?

I can't really comment on how it might work in a live environment. If there are performers on stage monitoring the same audio or even experiencing the sound 'slap' of a live venue, 40-50ms is probably pretty hefty.

If it is more a case of mixing playback into a space, people do that all the time with Atmos without issue. I have mixed numerous films in larger theatrical mix stages using Dolby Atmos Bridge and have never once noticed any kind of detrimental latency. A DAB based workflow would be closer to 90-100ms.